252
L. S. Alaimo and P. Conigliaro
Fig. 2 Distribution of emotional status affects: nervous; down; calm; depressed; happy.
N = 15,354
(2017), we can consider poset evaluation as a way to derive a complete order out
of a partial order: once the evaluation scores have been assigned to the elements of
a poset, they can be ordered in a linear order. Through a simple procedure (Fattore
2017), it is possible to assign to each element of a finite poset a score representing
its position in a “low-high” axis. In this way, we obtain the average rank, i.e. the
average position assumed by a profile in each of the considered linear extensions.
We cannot use the average rank – representing itself a synthesis – as an expression
of the level of ES, as it is a measure of (average) positioning of the profile in the
general order. Furthermore, if we apply the quantile criteria to split the profile into
groups, we could find in a single group profiles expressing very different levels
of ES and profiles of the same level in contiguous groups. In this way, we risk
committing errors in attributing an ES level to individual respondents. In addition to
the position in the general order, we need to take into account the situation in terms
of ES deprivation of each profile, by establishing a criterion to assign the response
profiles to the “deprived” or “not-deprived” category. We can do this by defining a
threshold profile. This choice allows characterizing the distribution according to a
conceptual definition of the phenomenon (it may contemplate the meaning attributed
to the item in a particular culture or other considerations). Starting from the partial
order among profiles, the aim of our analysis is to identify the deprived profiles, with
respect to the threshold, in each linear extension of the poset (Fattore 2016). For the
ES, we have defined as threshold profile the combination [2,2,3,2,3] 15 . It identifies
15 Other thresholds could also be added to better characterize and distinguish profiles. However, in
this case we have not identified another suitable point of discrimination at conceptual level.
L. S. Alaimo and P. Conigliaro
Fig. 2 Distribution of emotional status affects: nervous; down; calm; depressed; happy.
N = 15,354
(2017), we can consider poset evaluation as a way to derive a complete order out
of a partial order: once the evaluation scores have been assigned to the elements of
a poset, they can be ordered in a linear order. Through a simple procedure (Fattore
2017), it is possible to assign to each element of a finite poset a score representing
its position in a “low-high” axis. In this way, we obtain the average rank, i.e. the
average position assumed by a profile in each of the considered linear extensions.
We cannot use the average rank – representing itself a synthesis – as an expression
of the level of ES, as it is a measure of (average) positioning of the profile in the
general order. Furthermore, if we apply the quantile criteria to split the profile into
groups, we could find in a single group profiles expressing very different levels
of ES and profiles of the same level in contiguous groups. In this way, we risk
committing errors in attributing an ES level to individual respondents. In addition to
the position in the general order, we need to take into account the situation in terms
of ES deprivation of each profile, by establishing a criterion to assign the response
profiles to the “deprived” or “not-deprived” category. We can do this by defining a
threshold profile. This choice allows characterizing the distribution according to a
conceptual definition of the phenomenon (it may contemplate the meaning attributed
to the item in a particular culture or other considerations). Starting from the partial
order among profiles, the aim of our analysis is to identify the deprived profiles, with
respect to the threshold, in each linear extension of the poset (Fattore 2016). For the
ES, we have defined as threshold profile the combination [2,2,3,2,3] 15 . It identifies
15 Other thresholds could also be added to better characterize and distinguish profiles. However, in
this case we have not identified another suitable point of discrimination at conceptual level.